Detection of coherent infrasonic signals  in the Fairbanks and Antarctic array data sets is done both visually, on the monitor screen at the Geophysical Institute hub site, and also by  computer-analysis of each UT day of infrasonic data. If a significant infrasonic wave train is seen in the un-filtered data on the computer-monitor screens at the hub, the operator can then use a software program called Datascan to determine the wave train’s time of arrival, its coherence across all 8 channels, its horizontal trace velocity, its azimuth of arrival, as well as its spectral content and a measure of whether or not the signal was due to a plane-wave traversing the array. The Datascan program software is supported by either Matlab 6.0 or Matlab 7.0 platforms. Further information about the Datascan program for infrasonic array data analysis can be obtained from Charles Wilson, John Olson. or Curt Szuberla

The infrasonic microphone pressure data is digitized at 20 samples per second at both the I55US and I53US infrasonic arrays. Automatic computer analysis of the I55 and I53 data sets is done for each UT day of data using three separate frequency pass bands: for long period signals [ 0.015 to 0.10 Hz], for the microbaroms channel [ 0.1 to 0.5 Hz], for short period signals [1.0 to10 Hz]. A sliding window of selectable length is used in the automatic analysis of the infrasonic data. The output of this computer analysis consists of “mat files “that contain the trace velocity (Vt), azimuth (Az), coherence (MCCM), and time (t) for each successive data window throughout the 24 hour UT day. There are also “tif files” in the output that are graphical displays of the computer analysis for each UT day displaying the time history of the following parameters; F-Statistic, MCCM, Vt and Az for each successive analysis window for the day. Figure 1 shows a sample figure of a typical graphical display from the computer analysis for day 194 of 2005 at I53US.

Figure 1. Typical data showing day 194 of 2005 at I53US.

The band pass filter frequency limits are given at the top of the figure together with the Sample Rate for the data, the Sample Window size and the Update size. For a sample window of 10000 points and an update size of 5000 points the overlap in time is just one half the analysis window. The top two histograms in the figure show the Probability Density Function statistics of MCCM and the F-Statistic. The MCCM value is the Mean of the Cross-Correlation Maxima, normalized to unity, as  determined from the microphone time-series pressure data, for  each of the 28 separate pairs of microphones in the array of 8 sensors. The F-Statistic is a measure of the signal to noise ratio for coherent signals as determined from the pressure time-series data. Graphs are given of the values, from each analysis window throughout the day, for F-Stat, MCCM, Vel, and Azim versus time in hours. For the F-Stat graph the points shown in green are for all those data windows for which F-Stat is greater than 5. The F-Stat graph is auto-scaled in ordinate value to the maximum value for the day. In the MCCM graph the data points are shown by a blue x if it is less than the selected threshold value of 0.6 and by a red circle if their value is above the threshold. The number of red circles in the MCCM graph is listed as the number of “Events”. In the velocity graph the bottom ordinate on the graph is 0.10 Km/sec therefore any data window that has a velocity less than 0.10 Km/sec will not show on the graph. For both the velocity and the azimuth graphs the red circles represent analysis windows for which MCCM was greater than 0.6 in value. At the very bottom of the figure is a panel that represents, by a red circle, the presence of a plane-wave detected in the analysis window for which MCCM was greater than 0.6. A blue x in this panel is for a plane-wave in the analysis window for which MCCM was less than 0.6. At the bottom of the figure in the right hand corner is a string of eight numbers, either a zero or a one, for each microphone that was used in the data analysis. Thus the string 11011111 would mean that only sensor number 3 was not used in the analysis.   The particular infrasonic signals detected at I53US between 10 and 20 UT, as shown in the figure, were high trace-velocity wave-trains from the north  that were probably associated with auroral activity.

At I53US and I55US there are thousands of infrasonic signal detections found each year by the daily MCCM computer analysis system. The natural sources from which we observe infrasound are: geomagnetic and auroral substorms, volcanic eruptions, mountain-generated wind turbulence (MAW), large earthquakes, avalanches in the mountains, marine storm-sea surface interactions, and bolides. The observed man-made infrasound is from:  mining excavation blasts, rocket launches, aircraft sonic booms, cannon fire and bomb testing, airplane crashes, and local aircraft operations. Discussions of these disparate infrasound sources can be found in the links to the left.